摘要
相关反馈技术是近年来图像检索中的重要研究方向,它有效地缩短了用户高层语义和图像底层视觉特征的差距,大大提高了系统的检索精度,SVM因其通用性和出色的分类能力逐渐被引入到图像检索系统中。为了进一步提高检索效率,采用三级反馈机制引入模糊相关,用户对检索结果标记为相关图像、模糊相关图像和不相关图像,并对经典的查询向量点移动算法进行修改,在此基础上运用多分类SVM提出一种新的相关反馈图像检索方法。试验表明这是一个有效的方法,提高了图像检索效率。
Recently, the relevance feedback technique has been one of the important research facts in CBIR. Because it has greatly reduced the gap between the high level notion and low level visual features, the retrieval results are better, because of its versatility and splendid classified ability, SVM are introduced gradually in the image retrieval system. To further raise the retrieval efficiency, use the third - level feedback mechanism introducing fuzzy relevance,users mark the result for the related image, the fuzzy related image and the non-correlated image, and revise the inquiry vector migration algorithm, based on this utilize multi - classification SVM to propose one new relevance feedback image retrieval method. Through experiments can see this is an effective method, raising the image retrieval efficiency.
出处
《计算机技术与发展》
2009年第8期65-68,共4页
Computer Technology and Development
基金
上海市教育科研基金项目(教05-31)
关键词
图像检索
相关反馈
模糊相关
SVM
多分类SVM
image retrieval
relevance feedback
fuzzy relevance
SVM
multi- classification SVM
